Loss-of-function (LoF) mutations in human protein-coding genes are known to play a major role in severe diseases such as cystic fibrosis and muscular dystrophy, and have also recently been shown to influence the risk of complex diseases such as type 1diabetes and Crohn's disease. We have recently conducted the largest systematic survey to date of human LoF variants, as part of the 1000 Genomes Project, which has confirmed the value of LoF variants for human disease studies and also identified key chalenges for the detection and interpretation of these variants. We propose to overcome these challenges by constructing robust, accurate tools for the annotation, characterization and high-throughput genotyping of LoF variants. Firstly, we will develop an integrated informatic pipeline (Annotation of LoF Transcripts, ALoFT) for the identification and filtering of all classes of LoF variant, including single nucleotide substitutions (SNPs), insertions and deletions. Secondly, we will exploit data from RNA sequencing experiments and disease mutation databases to create more accurate predictive models of the efects of genetic variants on gene expression and splicing, and of their probability of disease causation. Finally, we will apply the ALoFT pipeline and the predictive models described above to over 30,000 human exomes and genomes sequenced as part of other NIH-funded projects, using the resulting annotation as the basis for a publicly accessible database of validated LoF variants, dbLoF. We will use our functional annotation to develop a weighted association test and apply this to the discovery of novel disease risk variants in these sequenced individuals. In addition, we will use the catalogue of LoF variants identified in these samples to design a custom genotyping array permitting rapid, cost-effective interrogation of the majority of common human LoF variants in human cohorts, allowing the phenotypic effects of these variants to be assessed in separately funded association studies. This study will provide powerful tools for discovering and characterizing natural loss-of-function variants, and for exploring their potential association with human disease risk.